Predictive Models for Motor Outcomes From Deep Brain Stimulation in Parkinson's Disease: A Systematic Review

Area of Science:

• Neurology
• Neurosurgery
• Medical Informatics

Background:

• Deep brain stimulation (DBS) offers motor symptom improvement for Parkinson's disease (PD) patients, but outcomes vary significantly.
• Accurate pre-surgical prediction of motor outcomes is crucial for clinical decision-making and managing patient expectations.

Purpose of the Study:

• To systematically review and assess existing predictive models for motor outcomes following DBS in Parkinson's disease patients.
• To identify key data types, model limitations, and future directions for developing clinically applicable predictive tools.

Main Methods:

• Systematic literature search of Web of Science, PubMed, and Scopus for studies predicting pre-operative DBS motor outcomes in PD.
• Inclusion criteria focused on models using pre-surgical data and reporting motor outcomes.
• Analysis of 19 identified studies regarding data types, accuracy, generalizability, and prediction scope.

Main Results:

• Studies relying solely on clinical data show limited accuracy but possess higher statistical power and generalizability.
• Models incorporating neuroimaging data (e.g., MRI) may achieve higher accuracy but are often under-powered.
• Current predictions are primarily focused on short-term outcomes (first year), subthalamic nucleus targets, and overall motor scores.

Conclusions:

• There is a need for robust, generalizable predictive models for DBS motor outcomes in Parkinson's disease.
• Future models should incorporate diverse data types, predict multiple motor domains over time, and include confidence intervals.
• Publicly sharing models and datasets is essential for validation and clinical translation to improve patient care worldwide.